When Does Syntax Mediate Neural Language Model Performance? Evidence from Dropout Probes

Anonymous

Paper Link: https://openreview.net/forum?id=CqvI1aRa0-

Paper Type: Long paper (up to eight pages of content + unlimited references and appendices)

Abstract: Recent causal probing literature reveals when language models and syntactic probes use similar representations. Such techniques may yield ``false negative'' causality results: models may use representations of syntax, but probes may have learned to use redundant encodings of the same syntactic information. We demonstrate that models do encode syntactic information redundantly and introduce a new probe design that guides probes to consider all syntactic information present in embeddings. Using these probes, we find evidence for the use of syntax in models where prior methods did not, allowing us to boost model performance by injecting syntactic information into representations.

Copyright Consent Signature (type Name Or NA If Not Transferrable): Mycal Tucker

Copyright Consent Name And Address: Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139 USA

Presentation Mode: This paper will be presented in person in Seattle